more on trac

John Gilmore gnu at
Fri Nov 20 19:47:19 EST 2009

> I don't wish to bring forward all the old bugs, that would increase the
> bug load insanely.  (It would make the current 1.5 work harder because
> of "noise" bugs.)

Yeah, what he said!  If the team didn't fix it in a previous release,
it is never worth fixing.  We shouldn't track whether it's still a
bug, or test whether it's biting our new products.  It's not even a
bug any more.  It's now deemed to be a feature.  We have an insane
level of features now.</sarcasm>

I've never used a bug tracking system -- except one -- that wasn't
full of people inappropriately closing out bugs.  The Fedora and
Ubuntu bug tracking systems have special volunteer "QA" positions who
do nothing but, and robots that go through and close out every bug
reported against every prior release!  The "except one" system (PRMS)
did not allow anyone, other than the person who reported the bug, to
close it out.  It didn't matter how many releases went past with the
bug unfixed, it stayed in the system.  It was a hassle nagging the
customers to close the fixed bugs, but no bug was ever closed without
the customer agreeing it was fixed.  Cygnus wrote PRMS, and I insisted
on this feature.  Customers were paying us hundreds of thousands of
dollars for high quality support of free software, and we damn well
wanted to provide it.  When we thought it was fixed, we put the bug
into "Feedback" state and got the customer to tell us whether *they*
too thought it was fixed.  The customer could move it from Feedback
into Closed -- or back to Open if what we'd done didn't really fix
their problem.

I've basically stopped reporting bugs to Fedora and Ubuntu, because
why waste an hour of my time trying to improve the world that way,
when some idiot will just close it out without the relevant engineers
ever seeing or fixing it?  If OLPC wants users to spend the time
reporting bugs in trac, it should retain those bugs in trac until they
are actually fixed.  Else why not just dash off a note to the devel
mailing list instead?  Or merely live with the bug?  If a bug report
falls in a forest and nobody hears it, why bother?

I know, Quozl isn't proposing that we close out all the old bugs; he's
just proposing that we arrange our "milestone" reports so that nobody
will ever look at them.  I'm not clear on why this is an improvement.


More information about the Devel mailing list